The dense retriever should work out-of-the-box by providing a multilingual model, such as this one sentence-transformers/distiluse-base-multilingual-cased-v1, or a model specifically trained for Chinese.
I think the sparse retriever works if you provide custom stemmer and tokenizer (if needed).
A Chinese stop-word list is already provided.
AmenRa
commented
1 year ago
Hi,
The dense retriever should work out-of-the-box by providing a multilingual model, such as this one
sentence-transformers/distiluse-base-multilingual-cased-v1, or a model specifically trained for Chinese.I think the sparse retriever works if you provide custom stemmer and tokenizer (if needed). A Chinese stop-word list is already provided.
Let me know if you have any luck!
Elias